Patients with CTS who were admitted to the Neurology Outpatient Department of our hospital between March 2018 and January 2019 were recruited consecutively. Eligible patients were initially screened by one of the authors based on detailed history-taking and careful neurological examinations. The inclusion criteria were as follows: 1) the presence of numbness, tingling, or pain in at least two fingers from the thumb (digit 1 [D1]) to digit 4 (D4); 2) symptoms worsening at night (sleep) or during hand movements; 3) clumsiness, weakness, or muscle atrophy in the thenar region; 4) presence of unilateral hand symptom or much worse hand in bilateral symptomatic hands, 5) age > 20 years, and 6) symptoms lasting for > 2 months, even if they occurred intermittently. Provocative maneuvers (Tinel’s and Phalen’s signs) were recorded in all patients, but they were not required for a patient to be enrolled. The exclusion criteria were as follows: 1) the presence of a medical illness that can contribute to CTS (e.g., diabetes mellitus, thyroid disease, chronic liver or kidney disease, or alcohol abuse), 2) the presence of another peripheral nerve disorder (e.g., polyneuropathy, plexopathy, or radiculopathy), 3) previous injury or surgery of the symptomatic hand, 4) the presence of peripheral vascular disorders, or 5) pregnancy or breastfeeding. Written informed consent was obtained from all patients, and all of them underwent NCSs. The control group consisted of healthy subjects without evidence of clinical or electrophysiological CTS. Note that the hands of the enrolled patients that had never experienced any sensory problems and the hands with normal NCS results served as controls along with the included healthy subjects. This study was approved by the local ethics committee of our institution (approval no. GDIRB2017-360).

All patients completed the following patient- and physician-based questionnaires to assess their clinical severity: the symptom severity scale (SSS) and functional status scale (FSS) of the Boston Carpal Tunnel Questionnaire8 and the Historical-Objective (Hi-Ob) scale9. The SSS and FSS consisted of 11 and 8 items, respectively, and were rated from 1 to 5 points, with higher scores indicating greater pain and disability. The SSS and FSS scores were calculated as the unweighted mean of all answered items for the corresponding scale; that is, the sum score for the responses divided by the number of items. The maximum and minimum scores on the SSS and FSS were 5 and 1, respectively. The Hi-Ob scale was assessed by the same trained neurologist. If the patients had bilateral CTS symptoms, they were instructed to complete questionnaires for each hand. Both the patient and the examining physician were blinded to the NCS results.

NCSs were performed using standard techniques of surface electrical stimulation and recording and an electromyography (EMG) system (Nicolet Synergy, Natus, Middleton, WI, USA). Skin temperature was maintained at above 31.0°C. Motor, sensory, and mixed NCSs of the bilateral median and ulnar nerves were performed to reveal the possibility of polyneuropathy and asymptomatic CTS. Needle EMG examinations were performed of the median nerve and the muscles innervated by the C6 and C7 roots, depending on patient tolerance. For the median motor NCS, compound muscle action potentials (CMAPs) were recorded in the abductor pollicis brevis muscle for stimulation at approximately 5 cm from the wrist. Normal values were defined as ≥ 5 mV for CMAP amplitude and ≤ 3.9 ms for distal motor latency (DML). For the median sensory NCS, sensory nerve action potentials (SNAPs) were recorded at the wrist with stimulation 1) at the palm (8-9 cm from the recording electrode) and 2) at the index finger using the orthodromic method. The sensory nerve conduction velocity (SNCV) was determined from the peak latency and calculated in meters per second. Normal values were defined as ≥ 10 μV for the SNAP amplitude, ≥ 34 m/s for the palm-wrist segment, and ≥ 40 m/s for the index finger-wrist segment. Electrophysiological CTS was diagnosed according to the guidelines of the American Association of Neuromuscular and Electrodiagnostic Medicine.10 Median-ulnar sensory conduction was compared between the wrist and the ring finger, and considered abnormal when the latency difference was ≥ 0.5 ms if conventional median sensory conduction was normal in both segments.11 The NCS results were graded according to the Padua scale for severity on each hand.12

Because the numbers of scores on the Hi-Ob and Padua scales differed and the group with moderate CTS was much larger than the other groups (Table 1), we categorized those scores into two and three groups, respectively: 1) the Hi-Ob scale was categorized into class 1 (stages 1 and 2) for paresthesia and class 2 (stages 3, 4, and 5) for objective sensory or motor deficits, and 2) the Padua scale was categorized into class 1 (grades 1 and 2), class 2 (grade 3), and class 3 (grades 4 and 5) depending on whether NCSs revealed abnormal SNCV only, abnormal DML, or abnormal SNAP and CMAP amplitudes, respectively.

The skin temperature of both hands was measured using a digital infrared thermographic imaging system (IRIS-XP, Medicore, Hanam, Korea) in an air-conditioned, window-less laboratory room that was maintained at 20-22°C and a humidity of 50-60%. The patients underwent IRT at least 24 hours after the NCS. They were prepared for IRT at 24 hours before the examination, with physical therapy, acupuncture, analgesics, and any topical agents affecting the skin not allowed during that time. On the day of the test, the patients were instructed to remove their top and remain in the room for 15 minutes to equilibrate. The temperatures at 14 sites (bilateral D1 to digit 5 [D5] and the bilateral thenar [Th] and hypothenar [Ht] regions) were measured as the regions of interest (ROIs) (Fig. 1). Because we included patients with bilateral CTS with a profoundly symptomatic single hand, we calculated the absolute temperature differences (ΔT-values) between each two of the seven sites in one hand rather than the thermal asymmetry between the left and right corresponding sites.

With reference to a previous study,13 we determined D1, D2, D3, and Th as the locations of median nerve vasomotor innervation, and D5 and Ht as the location of ulnar nerve vasomotor innervation. D4 was excluded from the analysis for clarity since both median and ulnar nerve innervated in D4. For sensitivity and specificity calculations, a significant difference was regarded as being due to the carpal tunnel lesion when one of ΔT(D1-D5), ΔT(D2-D5), ΔT (D3-D5), ΔT (Th-Ht) exceeded the reference value and simultaneously ΔT (D5-Ht) was below the reference value (0.4°C14, 0.6°C15, and 1.0°C16, respectively). The median index (MI) was calculated as (D1-D2) + (D2-D3) + (D1-D3),17 and the median-ulnar difference (MU) was calculated as (D1-D5) + (D2-D5) + (D3- D5) + (Th-Ht).18 Higher MI and MU values mean that the skin temperature in the median nerve distribution area is more variable, and this might reflect a malfunction in sympathetic vasoregulation.

Nonparametric tests were performed because Kolmogorov-Smirnov tests indicated that not all of the data conformed to a normal distribution. Continuous variables were examined using Mann-Whitney, Kruskal-Wallis, and univariate linear regression tests, as appropriate. p-values of less than 0.05 were considered to indicate statistical significance, and Bonferroni correction was applied for multiple comparisons. The sensitivity and specificity of NCSs, IRT, and their combination in diagnosing clinical CTS were calculated. Correlations between the SSS, FSS, Hi-Ob scale, Padua scale, and IRT results were examined using Spearman’s test. All statistical analyses were performed using Statistical Package for the Social Sciences (version 21.0, IBM Corp., Armonk, NY, USA).

A side-to-side comparison of hand temperatures at the seven ROIs in eight healthy controls showed similarity and high correlation coefficients for each comparison, confirming the presence of symmetry in hand temperatures (Supplementary Table 1). When comparing healthy controls (n = 8) and patients with CTS who showed clinical symptoms in total agreement with NCS results (n = 14), absolute side-to-side differences in the D2 temperature tended to be greater in patients with CTS (mean value, 0.71°C vs. 0.26°C), though the difference did not reach statistical significance (p = 0.266) after adjusting for age, sex, and BMI. Considering all hands (n = 66), regardless of NCS results, comparisons of regional ΔT in the same hand between symptomatic CTS and healthy hands did not reveal any differences. However, regardless of symptoms, hands with abnormal NCS results showed higher MI than hands with normal NCS results (mean value, 0.86°C vs. 0.54°C), but this difference also did not reach statistical significance (p = 0.055) after adjusting for age, sex, and BMI (Supplementary Table 2).

Based on a pure clinical diagnosis of symptomatic and asymptomatic hands, the sensitivity of IRT decreased and its specificity increased as the cutoff value increased (Table 2). The sensitivity and specificity of IRT were not superior to those of NCS. Combining the existence of either IRT (cutoff value of 1.0°C) or NCS abnormalities resulted in a modest increase in the sensitivity from 91.4% to 94.3%.

Regarding the 32 hands with CTS symptoms and NCSproven CTS, ΔT (Th-Ht) tended to be higher for class 2 on the Hi-Ob scale (p = 0.085), while ΔT (D3-D5) tended to be higher for class 3 on the Padua scale (p = 0.106), which were considered more-advanced stages. Skin temperature was lower in hands with moderate-to-severe CTS than in hands with mild CTS (Fig. 2).

The SSS and FSS scores showed a strong mutual correlation (Spearman’s rho, 0.882; p < 0.001). However, they did not show a correlation with the Hi-Ob scale, Padua scale, and IRT parameters (ΔT between ROIs, MI, and MU) in 32 hands with CTS symptoms and NCS-proven CTS. The average SSS and FSS scores did not differ significantly with the classes on the Hi-Ob and Padua scales (Supplementary Fig. 1, 2). We further conducted a correlation analysis between SSS subscores concerning pain (items 1-5) and IRT parameters, which revealed no correlation. However, 11 hands with high SSS scores (> 4.0) showed a positive correlation with ΔT (Th-Ht) (Spearman’s rho = 0.812; p = 0.177) after adjusting for age, sex, and BMI, and hands with CTS with higher SSS scores tended to be colder in the thenar region (Fig. 3).

We reviewed 11 hands with a discrepancy between symptoms and NCS results. The mean SSS and FSS scores for hands with CTS symptoms (n = 3) were comparably high, at 3.18 and 2.75, respectively. The regional ΔT of the ROIs in 11 hands did not show abnormal differences in the territory of the median nerve. However, the reliability of the interpretation was limited by the small number of samples available for inclusion in the statistical analysis.